A Type-2 Fuzzy Intelligent Agent Based on Sparse Kernel Machines for Reducing Bullwhip Effect in Supply Chain

Abstract

In this paper, a new type-2 fuzzy intelligent agent system (T2F-IAS) for reducing bullwhip effect in a supply chain is proposed. This system uses a special kind of sparse kernel machines, called support vector regression, for forecasting future demands of each echelon in a supply chain. The T2F-IAS includes a data collector agent and a rule generator agent. A type-2 fuzzy c-regression clustering model is employed in the rule generator agent for generating the most proper rules. This agent uses an interval type-2 fuzzy (IT2F) hybrid expert system for demand forecasting. Moreover, adaptive network based fuzzy inference system (ANFIS) is applied to learn parameters used in the agents. Thereafter, the results of the proposed T2F-IAS are compared with type-1 fuzzy intelligent agent system (T1F-IAS) and a method in literature for validating the proposed method. The results indicate that bullwhip effect and forecasting error are remarkably reduced by using the proposed T2F-IAS.